PyPy buildbot times out

[Sean Farley]

Joshua Root writes:

> On 2013-5-19 07:50 , Sean Farley wrote:
>> On Sat, May 18, 2013 at 4:39 PM, Ryan Schmidt <ryandesign at macports.org> wrote:
>>>
>>> On May 18, 2013, at 12:34, Sean Farley wrote:
>>>
>>>> I have a patch to fix pypy Using The Right Compiler but noticed that the buildbot was failing with a timeout error. Is there any way to increase the timeout for pypy?
>
> No, the timeout value is global. I don't think we want to wait any
> longer for builds that are actually stuck to be killed.

Yeah, that's probably a good idea.

>>> How long is it taking? Why is it timing out? Is it actually compiling when it times out, or is it stuck somewhere? When I update the php port the buildbot takes hours to build, but doesn't time out.
>> 
>> Hmm, that's a good point. It seems that the job is killed after 1200
>> of no output. I'll have to try to build this on my laptop and see.
>
> It's timing out in the C generation or the final link step after
> compiling the C. I suspect that the slaves may be thrashing, as the
> build takes a lot of memory. I don't think it ever took anywhere near 20
> minutes between printing lines when I built locally (with 8 GB of RAM).
>
> If it's possible to allocate more memory to the slave VMs, that might help.

I have a log file with each line timestamped (perhaps I could create a
ticket and upload the log?) and the longest time between two lines is
168s so I guess it's believable that the VMs could be thrashing around?
I can't say for sure without more details.

As for the patch I have, should I just commit it and if the build fails
again, ask Bill for some input?